102 CHAPTER 6. COMPOSITE MEDIA BASED ON AU/LIF ARRAYS0.250.20T modeL mode0.350.300.25T modeL modeR S0.15R S0.200.150.100.10400 600 800 1000a. [nm]b.400600 [nm]8001000Figure 6.5: Experimental (empty circles) <strong>and</strong> computed (continuous lines) R S spectra,with plane <strong>of</strong> incidence parallel (black curves) <strong>and</strong> perpendicular (red curves) the LiFridges. Left panel: 2D <strong>arrays</strong> <strong>of</strong> gold NPs on nanopatterned LiF(110) (Λ = 35 nm,t Au ≈ 5 nm, AFM data in fig. 6.2(a, b)). Right panel: same system after the deposition<strong>of</strong> a monolayer <strong>of</strong> Fe 3 O 4 /OA NPs (AFM data in fig. 6.2(c)).monolayer <strong>of</strong> deposited NPs, the first one was to increase the thickness <strong>of</strong> the effectivelayer by an amount equal to the mean diameter <strong>of</strong> the NPs, i.e. ≈ 17 nm. The secondcorrection was to rewrite the dielectric constant ε h <strong>of</strong> the host including the contribution<strong>of</strong> the magnetite NPs; we considered the gold NPs embedded in an effective host composedat 50% by LiF (with dielectric constant ε s ) <strong>and</strong> at 50% by a mixing <strong>of</strong> air <strong>and</strong> bulkmagnetite (with dielectric constant ε Fe3O 4, assuming that the dielectric constant <strong>of</strong> Fe 3 O 4nanoparticles does not differ from the bulk):ε h = 1 2 ε s + 1 2 (f ε Fe 3O 4+(1−f)ε air ) (6.1)where ε air = 1 <strong>and</strong> f is the magnetite filling factor. The latter is calculated by simplegeometrical considerations: assuming the Fe 3 O 4 /OA nanoparticles arranged on a squarelattice, <strong>and</strong> in contact with one another, we find one particle per volume <strong>of</strong> d 3 hydro (d hydrois the hydrodynamic size), while the volume <strong>of</strong> the magnetite core is 4π/3(d core /2) 3 (d coreis the size <strong>of</strong> the NPs core); the filling factor f is given by the ratio <strong>of</strong> the two volumes<strong>and</strong>, substituting the values obtained from TEM <strong>and</strong> DLS measurements, results f ≈ 0.3.The calculated R S spectra for the Au NPs array after the Fe 3 O 4 /OA deposition arereported in fig. 6.5(b). Despite the simplifications in treating the magnetite NPs layer,which probably determined the overestimation <strong>of</strong> the absolute values <strong>of</strong> reflectivity, thepositions <strong>and</strong> widths <strong>of</strong> the LSPs were reproduced in good agreement with the experiments,resulting <strong>of</strong> λ ∗ L = 645 nm <strong>and</strong> Γ∗ L = 223 nm for the L mode <strong>and</strong> λ∗ T = 576 nm<strong>and</strong> Γ ∗ T = 145 nm for the T mode. In particular, this also supports the consideration thatonly one monolayer <strong>of</strong> Fe 3 O 4 /OA NPs has been deposited.Inconclusion,wehaveshownthatthe<strong>self</strong>-<strong>organized</strong>Au/LiFsystemscouldbefruitfullyexploited as templates for the fabrication <strong>of</strong> more complex composite structures, that canatleastpartlyretainthemorphologicalcharacteristics<strong>of</strong>theoriginalsystem<strong>and</strong>itsopticalresponse, adding to these novel functionalities, like magnetic response in this specific case.
ConclusionsIn this thesis, we have discussed the fabrication <strong>of</strong> <strong>self</strong>-<strong>organized</strong> 2-dimensional <strong>arrays</strong> <strong>of</strong>gold nanoparticles, with independently tunable size, shape <strong>and</strong> periodic arrangement, <strong>and</strong>the characterization <strong>of</strong> their collective <strong>plasmonic</strong> response.The <strong>arrays</strong> were fabricated employing <strong>self</strong>-<strong>organized</strong> nanopatterned LiF(110) substratesas templates for guiding the particle formation. First, a regular ridge-<strong>and</strong>-valley(ripple) structure is spontaneously induced by means <strong>of</strong> homoepitaxial growth <strong>of</strong> LiF onLiF(110), with a variable periodicity Λ determined by the substrate temperature duringthe deposition. Typical values <strong>of</strong> Λ = 25÷60 nm by varying the temperature in the rangebetween 250 ◦ C <strong>and</strong> 450 ◦ C were found.Then, a thin layer <strong>of</strong> gold was deposited at incidence <strong>of</strong> θ = 60 ◦ , exploiting theshadow effect <strong>of</strong> the ripples ridges in order to form disconnected Au “nanowires”. Finally,the samples were mildly annealed at T = 400 ◦ C, promoting the dewetting <strong>of</strong> the goldnanowires into individual nanoparticles.After the dewetting, the surface <strong>of</strong> the samples was characterized by parallel chains <strong>of</strong>gold nanoparticles, regularly spaced according to the periodicity <strong>of</strong> the underlying ripples.The particles exhibited lognormal size distribution, with a typical st<strong>and</strong>ard deviationlower than 5 nm. Interestingly, we found a partial correlation also between the positions<strong>of</strong> particles belonging to adjacent ripples, <strong>and</strong> rectangular arrangements <strong>of</strong> Au NPs couldbe observed over large areas <strong>of</strong> the samples.By tuning the fabrication parameters, we showed the possibility <strong>of</strong> varying both thesingle NP mean size <strong>and</strong> in-plane aspect ratio <strong>and</strong> their mutual spacings in the <strong>arrays</strong>.The ripples periodicity Λ determined the NPs mean size <strong>and</strong> spacing in the directiontransversal to the LiF ridges, while the length <strong>and</strong> spacing along the ripples were relatedto the amount <strong>of</strong> deposited gold. Performing depositions under different conditions, wecould obtain several 2D <strong>arrays</strong> <strong>of</strong> gold NPs with in-plane aspect ratios between 1:1 <strong>and</strong>1:2, <strong>and</strong> typical dimensions in the 20÷50 nm range.The 2D <strong>arrays</strong> <strong>of</strong> gold NP were optically investigated by means <strong>of</strong> spectroscopic ellipsometry,reflectivity <strong>and</strong> transmissivity, in the visible <strong>and</strong> near-IR frequency range. Theoptical response <strong>of</strong> the <strong>arrays</strong> exhibited characteristic absorptions <strong>and</strong> reflectivity peaks,corresponding to the excitation <strong>of</strong> localized surface plasmons, i.e. collective oscillations <strong>of</strong>the electron gas confined inside the metallic NPs. Such resonances critically depended onsingle-particles parameters, like shape <strong>and</strong> size, <strong>and</strong> on extrinsic factors, like the dielectricenvironment or the inter-particles spacing. We investigated the position <strong>and</strong> width<strong>of</strong> the LSPs for different arrangements <strong>of</strong> the Au NPs, with particular emphasis on theoptical anisotropy between the longitudinal LSP mode, excited by an electric field alongthe ripples, <strong>and</strong> the transverse LSP mode, excited perpendicular to the ripples.In order to separately assess the contributions <strong>of</strong> the intrinsic <strong>and</strong> collective factorson the <strong>plasmonic</strong> response <strong>of</strong> the NP systems, we fabricated <strong>and</strong> discussed in detail two103